On Sun, 18 Jul 2010, Jan Van Boghout wrote:
I also don't think it should occur. Are you seeing it? The error code is used internally to create the simulated blocking API, so if one checks for an error code when there was in fact no error then you're pretty sure to get EAGAIN as the error code, since that was the last error libssh2 generated,
Oops, this was indeed a case of code blindness on my part. Everything is working as it should!
I've just pushed a change now that should make this situation less frequent, in that the internal _libssh2_wait_socket() function will now clear the error code. That should be the primary function that gets used after EAGAIN is set and used internally during blocking mode.
-- / daniel.haxx.se _______________________________________________ libssh2-devel http://cool.haxx.se/cgi-bin/mailman/listinfo/libssh2-devel
